Building a Protection Ecosystem: What Can We Learn from Recent Online Safety Challenges in Brazil

Content warning: discussion of self-harm and suicide.

How do we make protection arrive before the harm?

Sharing my thoughts publicly on such a sensitive topic, especially when it involves a tragedy and brings out such strong opinions, is always a challenge.

Even more so in writing. Text doesn’t carry tone of voice, intent, or context in the same way a conversation does. And when we’re talking about children, violence, platform responsibility, and the role of the government, any nuance that gets lost or misunderstood can come across as downplaying a tragedy or automatically and rigidly defending one side.

That’s why I tried to write this reflection from a place of care, without the intention of offering absolute answers. I hope it can contribute to a broader, less binary conversation about what we need to build from here. It’s an attempt to add to the conversation, not to end it.

With that in mind, the purpose of this blog post is not to defend a platform or question the need for regulation. And certainly not to take away from the responsibility of technology companies or the criminals themselves.

My intention is to reflect on the question that has become increasingly important throughout my work in online safety: how do we make protection arrive before the harm?

Without Losing Sight of What Brought Us Here

Before any discussion about Discord, ANPD, the ECA Digital, or regulatory measures, there is a very real tragedy at the heart of this entire debate.

In July, a 13-year-old girl died in Mato Grosso do Sul in a case that, according to the investigation, involved coercion and the encouragement of self-harm and suicide within an online community. We know she was not the only one, and that other cases have happened not only on Discord, but on other platforms as well.

The investigation eventually led to Operação Lívia (Livia Operation). According to Brazil’s Ministry of Justice and Public Security, the group used online spaces to target teenagers, subject them to psychological coercion, and encourage extreme acts of violence. Authorities also described a structured operation, with different roles for attracting victims, inducing violent behavior, and sharing the images that were produced.

It’s difficult to write about this without feeling the weight of what happened. I know this is not my story to tell, because I have never experienced anything close to the pain this family has gone through. But as a mother and as someone who works in online safety, this story, along with so many others we keep hearing about, hurts me and keeps me up at night.

Behind every discussion about regulation, moderation, platforms, or enforcement, there is a family that lost a child. No decision we make now can undo that loss.

That’s why I believe every question we ask, and every answer we give, needs to start with the gravity of what happened and the responsibility we have to prevent stories like this from happening again.

The ECA Digital: A Living Law, With Real-World Application

On Wednesday, August 12, this discussion entered a new chapter.

As a preventive measure, Brazil’s data protection authority, ANPD, ordered Discord to suspend its livestreaming feature in Brazil, as well as other functionally equivalent features such as video calls. Discord was given three business days to demonstrate full compliance with the suspension. The suspension will remain in place until the company demonstrates the implementation and effectiveness of technical, safety, and governance measures adequate to the risks identified, and ANPD expressly authorizes full or partial reinstatement of features.

On August 17, Discord published a public letter to its community in Brazil confirming that it is complying with the order. Screensharing and video calls are now unavailable to users in the country. Discord said it is working to restore access as quickly as possible, acknowledged the case that prompted the order, and said it has been and continues to collaborate with law enforcement.

It’s also important to say what did not happen: Discord was not blocked in Brazil. Servers, direct and group messages, voice-only channels, and other features that do not rely on video or screen sharing remain available.

The measure targeted the features that, according to ANPD’s technical assessment, presented robust evidence of posing the greatest risk to users under 18 in that context. The Agency also stated that the measures presented by the company so far were predominantly after the fact, meaning they came after harm had occurred, or had limited preventive effectiveness.

What happens next now that Discord has confirmed it is complying with the feature suspension?

From a practical standpoint, this situation deserves attention for two reasons.

First, the principles and obligations established by the ECA Digital translated into a concrete case. A risk was identified, an order was issued, a deadline was set, and the platform responded within it. That gives us a tangible example of how this relatively new law functions in practice.

Second, the regulatory response was preventive, targeted and proportionate to the risk identified. Rather than blocking the platform as a whole, ANPD temporarily suspended the functionality it considered most critical and made its return conditional on demonstrating adequate and effective safeguards are in place.

This practical application reinforces an important idea: safety cannot be measured simply by whether a policy, a report mechanism, or certain tools exist. We also need to assess whether those protections are effective when a risk moves from hypothetical to real.

The same focus on prevention and effectiveness should shape how we think about digital products more broadly: understand the risks associated with different experiences, build safeguards that are proportionate to those risks, measure their effectiveness, and keep adapting as risks evolve. This approach is closely aligned with the principles of Safety by Design and is also reflected in the ECA Digital, which requires protective measures to be considered from the design stage and throughout the operation of products and services likely to be accessed by children and adolescents.

But this decision and what follows from it, also leave us with an even harder question.

What Happens to These Groups Now?

Reducing a risk in a specific feature is an important part of the response. The next question is what happens to the people, relationships, and dynamics that existed in that environment from here on.

Experience shows that people involved in violent and abusive groups like these don’t simply disappear when restrictions are imposed, whether on a tool or an entire platform. These groups tend to reorganize, find another feature, move to another service, or seek out smaller, more private spaces that may have fewer safety and moderation resources or less cooperation with authorities. That is now no longer hypothetical. The features are off for Discord users in Brazil as of this week, and the people who used them are still here.

That’s not an argument against intervention. The possibility of migration should never be used as a reason for a platform to neglect its safety responsibilities or for a regulator to refrain from acting when faced with a serious risk.

But a necessary intervention should not be mistaken for the end of the problem.

The reality in Brazil itself helps show why. In the same week that it presented the results of Operação Lívia, the Ministry of Justice also detailed Operação Rede Interrompida (Interrupted Network Operation).

In this second case, an extremist network used Telegram communities to spread neo-Nazi, antisemitic, and misogynistic content, recruit new members, and discuss plans for violent acts. The investigation involved multiple states and brought together civil police forces, Ciberlab, the Public Prosecutor’s Office, and Abin (Brazilian Intelligence Agency).

These are separate investigations, and I’m not suggesting that the two groups are connected.

What they do show, however, is a broader pattern: this phenomenon does not belong to a single platform.

Violent networks use different services, languages, and mechanisms to recruit, create a sense of belonging, test boundaries, and organize actions.

Data from the Brazilian Public Security Forum shows that the digital environment surrounding youth violence deserves attention. A study conducted in partnership with Timelens found a 360% increase between 2021 and 2025 in posts containing threats against schools across the major platforms monitored. By May 21, 2025, more than 88,000 direct mentions of threats against students, teachers, and school principals had already been recorded. The study also found a significant increase in the share of comments glorifying perpetrators of attacks, from 0.2% in 2011 to 21% in 2025.

These findings are concerning, but they also require careful interpretation. A post does not equal a radicalized person, and an online threat does not necessarily mean there is a concrete intention to commit violence. Still, they suggest that we are not dealing with completely isolated incidents. There is a digital environment where threats, glorification of violence, and extremist content circulate, find an audience, and, in some cases, contribute to processes of recruitment and radicalization. The Brazilian Public Security Forum itself has highlighted the need to understand school violence as part of broader social dynamics, including those mediated by digital technologies.

Why is Another Person’s Suffering Collective Entertainment?

This broadens the discussion beyond a single platform. We need to understand how these networks form, recruit, and reorganize, but also why some young people find belonging and validation in them, while values like empathy, respect, and responsibility don’t always have the same power to engage them.

At the same time, it’s essential to strengthen our ability to identify and disrupt these networks and ensure that those who deliberately promote exploitation, coercion, and violence in these environments are held accountable.

This Is Also an International Challenge

Brazil is not alone in trying to respond to this kind of threat. In the United States, the FBI has been warning about violent online networks commonly referred to as 764, although there are numerous offshoot groups and other names associated with this phenomenon.

I’m not suggesting any connection between the Brazilian case and 764. The parallel here is in how these networks operate and how they exploit the vulnerabilities of children and teenagers.

According to the FBI, these groups seek out vulnerable young people on online platforms, build relationships of trust, and may gradually turn to manipulation, blackmail, and coercion to push them toward sexual or violent behavior, including self-harm and suicide attempts. The agency has also reported cases where members of these networks watched livestreams of self-harm and used the material to gain greater control over victims.

In February this year, the FBI said it was investigating more than 350 people connected to these networks in the United States, with all 56 of its field offices involved in the investigations. The agency also describes a similar pattern: many approaches begin on open platforms, and once a relationship has been established, victims may be moved to more private or encrypted communication spaces.

This doesn’t mean that all of these cases are the same. It does, however, show that we’re facing a challenge that crosses platforms, technologies, communities, and borders.

Online Safety Requires a Protection Ecosystem

This is where I believe we have a lot of room to broaden and deepen the conversation.

Online safety does not depend on a single tool, a single company, or a single regulatory decision. It requires a protection ecosystem in which different actors have clear roles, the right resources, and the ability to act in a coordinated way.

And perhaps that is also one of the reasons this problem has been so difficult to solve. It involves many parts of society and the digital ecosystem, with distinct responsibilities, different levels of information and decision-making, and risks that cut across platforms, institutions, and even borders.

A failure or lack of action at one point in this system can create a gap in protection. And coordinating so many roles, responsibilities and capabilities is far more complex than finding a single measure that can solve everything.

This protection ecosystem is built across different areas:

Product and platforms: Companies need to understand the risks associated with their features and incorporate Safety by Design into their products, including when those were not originally built with these principles in mind. That means creating age-appropriate experiences, providing effective reporting and response mechanisms, and continuously assessing whether those protections actually work in practice.

Regulation and enforcement: Regulators need the authority, specialized technical knowledge, and appropriate tools to investigate, require changes, and take action when risks, violations or serious safety concerns are identified.

Investigation and intelligence: Violent and criminal networks can move across services, accounts, states, and countries. This requires cooperation between authorities and the ability to follow these networks as they migrate between platforms or reorganize in more private spaces.

Justice system: Prosecutors and courts play an essential role in assessing conduct, enforcing the law, and holding those who commit crimes or cause harm accountable. An effective protection system must be able not only to identify risks and disrupt harmful behavior, but also to bring those responsible before the appropriate authorities so there are consequences where warranted.

This also raises the question of how effective individual accountability really is. What consequences do those who recruit, manipulate, coerce, or incite others to engage in violence actually face? And are those consequences effective enough to also contribute to prevention? This is a discussion that deserves deeper consideration, but it is also an integral part of this protection ecosystem.

Research and knowledge building: Academic institutions, research centers, and specialized organizations help understand how risks emerge, who is most vulnerable, which interventions work, and which may have unintended consequences. In Brazil, InternetLab, CGI.br/NIC.br, and different universities produce important data and research so that decisions about online safety are not based solely on perceptions or isolated incidents.

Civil society organizations: Organizations such as SaferNet Brasil, Instituto Alana, and other groups focus on the rights of children and adolescents, working across education, prevention, reporting, support channels, research, advocacy, and collaboration with authorities. Internationally, organizations such as the Tech Coalition, WeProtect Global Alliance, ECPAT International, and other centers focused on extremism and violent networks connect different countries and help develop responses to threats that cross borders.

Education, awareness, and training: Young people need the knowledge, skills, and support to recognize manipulation, coercion, blackmail, and the dynamics of violent groups. Families should be equipped with information and tools to stay involved in the digital lives of children and teenagers, without placing the entire burden of protection on them.

Schools require ongoing support to address digital citizenship and safety, critical thinking, well-being, and the importance of knowing when and where to seek help.

Community managers, moderators, developers, and teams responsible for building new games, apps, and digital communities also should also have access to training, best practices, standards, and safety tools. Not every company has robust online safety teams, and making this knowledge more accessible is also part of building an effective protection system.

Listening to and involving children and teenagers: If policies, products, and protection mechanisms are designed for children and teenagers, their experiences should also inform how those solutions are built. That means creating safe spaces to hear from them, understand how they perceive risk, how they seek help, and which tools actually make sense in their daily lives.

A survey released by Instituto Alana in August 2026 helps illustrate this point: 44% of the teenagers surveyed said they block accounts or content, and 37% said they report inappropriate profiles and content. At the same time, 71% said they believe the internet needs to improve, but only 19% felt they could do something to help make that change happen.

Mental health and support networks: Some of these networks deliberately target young people who are isolated, emotionally vulnerable, experiencing psychological distress, or looking for connection and belonging. Prevention therefore also means strengthening support networks outside platforms, expanding access to mental health support, and improving our ability to identify these signs before they can be exploited.

Communities and users: This is a role that still seems underestimated to me. Online communities are not just places where risks happen; they can also be part of the protection system. Users often notice concerning behavior, support one another, report content, and alert moderators when they believe someone may be at risk.

For that to work, however, people need to feel that speaking up will lead somewhere. Community managers and moderators need tools, processes, escalation pathways and support to act when a situation become serious. And users at risk need to be heard at the moment when timely intervention matters most.

Discord’s letter about suspending screensharing and video calling ends with a request to its Brazilian users: share your stories so that regulators and legislators can understand how the platform is really used and what it means to people and communities. That kind of participation can also be part of a broader conversation about how safety measures affect legitimate communities and how those communities can contribute to stronger protections.

And there is one dimension that cuts across and connects all of these areas: culture. Culture shapes what we value as a society, the boundaries we set, and how we respond to violent and abusive behavior.

We need more conversations, across different spaces and not in silos, about how violence, humiliation, coercion, and suffering could gradually become normalized within certain online groups, and what each actor can do, within their own sphere of responsibility, to interrupt these dynamics.

Even when we disagree, we need to preserve openness, listening, and curiosity so that this conversation can keep moving forward.

Complex Problems Require Systems Thinking

When we look at all these dimensions together, it becomes easier to understand why protecting children and teenagers requires such a broad and deep approach.

The complexity of the problem cannot be used to weaken platforms’ responsibility for user safety. But holding platforms accountable should not make us lose sight of the broader complexity of the problem.

There is no single point of intervention, just as there is no single actor capable of seeing or controlling the entire system. Each part of this ecosystem brings different information, capabilities, and limitations. No one actor can see the whole picture alone.

That interdependence is precisely what makes cooperation and coordination a necessity, not simply a choice or a one-off response.

It’s also not enough for each part of the system to work well in isolation. Its strength lies in how responsibilities, knowledge, and capabilities connect and complement one another. Shared responsibility does not mean diluted responsibility. It means connecting and complementing different responsibilities to create a real network of prevention, protection, and response. This principle is also reflected in the ECA Digital.

In that sense, ANPD’s decision and its ability to oversee and enforce the Digital ECA are important parts of this ecosystem, but the protection system does not end there. When a group migrates, a new threat emerges, or a risk changes shape, the system needs to be able to detect, learn, and respond in a coordinated way across platforms, institutions, and communities.

That, to me, is what it means to think about online safety as a protection ecosystem: not a single solution, nor a vague sense of responsibility that belongs to “everyone” and therefore risks belonging to no one in practice.

It means bringing different responsibilities, knowledge, and capabilities together around the same goal: reducing risks and making sure children and teenagers are protected before harm occurs.

A note on our work with Discord
GamerSafer is a Discord partner on the Player’s Guide and Wellbeing Principles, an initiative focused on helping communities strengthen player safety and wellbeing. We also build products for the communities that run Discord servers: GS Defender, a trust and safety bot, and DefenderNet, a cross-community threat intelligence network.

References

1. ANPD — Medida preventiva determina suspensão das transmissões ao vivo no Discord no Brasil. (ANPD - Preventive Measure Orders the Suspension of Live Streaming on Discord in Brazil). August 12, 2026. https://www.gov.br/anpd/pt-br/assuntos/noticias/em-medida-preventiva-anpd-determina-que-discord-suspenda-transmissoes-ao-vivo-no-brasil
2. Ministério da Justiça e Segurança Pública — Operações Lívia e Rede Interrompida. (Ministry of Justice and Public Security - Operations Lívia and Interrupted Network). August 4, 2026. https://www.gov.br/mj/pt-br/assuntos/noticias-1/operacao-livia-e-rede-interrompida-mobilizam-ministerio-da-justica-e-seguranca-publica-e-policias-civis-de-oito-estados
3. Ministério da Justiça e Segurança Pública — ECA Digital. (Ministry of Justice and Public Security – the ECA Digital.) March 20, 2026. https://www.gov.br/mj/pt-br/assuntos/sua-protecao/sedigi/eca-digital/eca-digital-1
4. Fórum Brasileiro de Segurança Pública / Timelens — Pesquisa sobre ameaças a escolas nas redes sociais. (Brazilian Forum on Public Security / Timelens - Study on Threats Against Schools on Social Media). June 11, 2026. https://fontesegura.forumseguranca.org.br/pesquisa-do-forum-brasileiro-de-seguranca-publica-aponta-crescimento-de-360-no-numero-de-postagens-com-ameacas-a-escolas-em-quatro-anos/
5. FBI — FBI Boston Warns of Nihilistic Violent Extremists Targeting Children and Vulnerable Victims Online. February 19, 2026. https://www.fbi.gov/contact-us/field-offices/boston/news/fbi-boston-warns-of-nihilistic-violent-extremists-targeting-children-and-vulnerable-victims-online
6. InternetLab — Pesquisa sobre hábitos, usos e impactos de plataformas de vídeos curtos para adolescentes. (InternetLab - Study on Teenagers’ Habits, Uses, and Experiences with Short-Form Video Platforms). July 11, 2024. https://internetlab.org.br/pt/noticias/internetlab-e-rede-conhecimento-social-lancam-pesquisa-sobre-habitos-usos-e-impactos-de-plataformas-de-videos-curtos-para-adolescentes/
7. Instituto Alana — Crianças e Adolescentes Primeiro: ECA Digital e proteção online. (Children and Teenagers First: the ECA Digital and Online Protection). 2026. https://alana.org.br/wp-content/uploads/2026/03/eca-digital-protecao-online-criancas-adolescentes-livro-digital.pdf
8. Discord — A letter to the Discord community in Brazil. August 17, 2026. https://discord.com/blog/a-letter-to-the-discord-community-in-brazil